Skip to content

feat(metax): route basic Llama through canonical InfiniOps #1498

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
wooway777 merged 1 commit into from
Aug 13, 2026
Merged

feat(metax): route basic Llama through canonical InfiniOps #1498

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion src/infinicore/nn/rope.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -78,7 +78,7 @@ void RoPE::initialize_cache() {
INFINICORE_NN_BUFFER_INIT(cos_cache, ({max_seq_len_, cache_dim}, dtype_, device_));

#ifdef ENABLE_INFINIOPS_API
if (device_.getType() == Device::Type::NVIDIA && !mrope_section_) {
if ((device_.getType() == Device::Type::NVIDIA || device_.getType() == Device::Type::METAX) && !mrope_section_) {
INFINICORE_NN_BUFFER_INIT(cos_sin_cache, ({max_seq_len_, rotary_dim_}, dtype_, device_));
}
#endif
Expand Down
4 changes: 3 additions & 1 deletion src/infinicore/ops/random_sample/random_sample.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,9 @@ namespace {
#ifdef ENABLE_INFINIOPS_API
bool tryGreedyWithInfiniOps(Tensor indices, Tensor logits, int topk) {
const auto dtype = logits->dtype();
if (logits->device().getType() != Device::Type::NVIDIA
const auto device_type = logits->device().getType();
if ((device_type != Device::Type::NVIDIA
&& device_type != Device::Type::METAX)
|| topk != 1
|| logits->ndim() != 1
|| logits->numel() == 0
Expand Down
5 changes: 4 additions & 1 deletion src/infinicore/ops/rotary_embedding/rotary_embedding_infiniops.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ void *plan(const Tensor &positions,
bool is_neox,
int64_t rope_dim_offset,
bool inverse) {
INFINICORE_ASSERT(query->device().getType() == Device::Type::NVIDIA);
INFINICORE_ASSERT(query->device().getType() == Device::Type::NVIDIA || query->device().getType() == Device::Type::METAX);
return new PlannedMeta{
TensorMeta(positions),
TensorMeta(query),
Expand Down Expand Up @@ -77,6 +77,9 @@ static bool registered = []() {
RotaryEmbedding::plan_dispatcher().registerDevice(Device::Type::NVIDIA, &plan);
RotaryEmbedding::run_dispatcher().registerDevice(Device::Type::NVIDIA, &run);
RotaryEmbedding::cleanup_dispatcher().registerDevice(Device::Type::NVIDIA, &cleanup);
RotaryEmbedding::plan_dispatcher().registerDevice(Device::Type::METAX, &plan);
RotaryEmbedding::run_dispatcher().registerDevice(Device::Type::METAX, &run);
RotaryEmbedding::cleanup_dispatcher().registerDevice(Device::Type::METAX, &cleanup);
return true;
}();

Expand Down
2 changes: 1 addition & 1 deletion xmake.lua
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -416,7 +416,7 @@ local function build_infiniops_external(xmake_os)
"-DGENERATE_PYTHON_BINDINGS=OFF",
"-DCMAKE_BUILD_TYPE=Release"
}
if has_config("nv-gpu") then
if has_config("nv-gpu") or has_config("metax-gpu") then
table.insert(cmake_config_args, "-DWITH_TORCH=ON")
table.insert(cmake_config_args, "-DINFINI_OPS_TORCH_OPS=argmax")
end
Expand Down
Loading